Create CLUT from film

Hi,

Haven’t been here for quite a while, I’m quite sorry. It was a pleasure to meet and shoot with you on LGM and I am still quite inspired by the presentations (and Pat bringing the softlighter with him on the flight).

This weekend I was on a photography flea market and got a good deal on a Rollei 35 SE (not as pretty as the 35 S but the batteries are easier to replace with something non-mercury based) and I have been collecting parts to get it working.

I also got myself some Agfa APX 100 (New), whose old version was identical to the Rollei 100 and the new version is probably reasonably close to the Rollei 80s which is by far my favorite film emulation in G’MIC. Now I’d love to contribute a CLUT for this film as well (though it already has one, now that I looked at it, but it does not say whether it is the old one or new one), but I don’t know what to do.

I have access to:

  • The Rollei 35 SE with a 5.6V meter which will have to work with 6V so it’ll probably overexpose
  • Agfa APX 100 New, which people say is rather ISO80
  • Access to some photography lab to develop it
  • A digital camera

But I have no idea how to go from this to some preset that can (hopefully) be included in G’MIC.

Any pointers would be appreciated!

1 Like

Hi Leonidas,

Have you tried searching pixls.us for the word CLUT?

I did, but the only thing I came up with was how to create a CLUT from a preset plus a lot of technical information on how LUTs work. Maybe I missed something.

The way I suppose it works is to photograph a known-color reference image (IT8 maybe) under known-color reference light and then get the image developed and scanned but I have no idea how to assure that developing and especially scanning be accurate? Or maybe I am thinking completely wrong.

Hello

When you hold a photo in your hand you are looking at the sum of all the stages it went through - how the film negative responded to the light in the camera, how the chemicals affected the film negative at the lab which developed it, how the negative was projected onto the blank photo paper, how the paper reacted to that, how the chemicals affected the paper, and how the end result looks in the light you’re viewing it under. If you’re thinking of scanning it, then your scanner’s response has a great impact as well.

As far as I can currently imagine, to create a Hald CLUT which changes colors in a digital image the way a given film stock would change them from their original values would be to shoot a color target (not an identity Hald CLUT image but a typical matte color target, the more patches the better), develop the film negative as carefully as possible, then scan the film negative using a calibrated and profiled scanner, match the exposures, then compare the Lab* values of this scanned film image vs what they originally were (Lab* values of color targets are given by the producer), then try to reproduce those differences (using curves) on an identity Hald CLUT image.

1 Like

Hi @Leonidas,

Hope you will find something of use here:
http://im.snibgo.com/ckbkClut.htm
http://im.snibgo.com/ckbkClut.htm#gimp
http://www.digicrea.be/hald-clut-film-emulation-with-gmic-plugin-for-gimp/
http://www.digicrea.be/2311-2/

There are some relatively affordable targets from Wolf Faust that might be appropriate. @houz may know which one is best for what you’re thinking.

Also pmjdebruijn over in #darktable might have some good workflow ideas. I’d imagine it’s as @Morgan_Hardwood mentioned, shoot a known color target, carefully develop, scan, and build out the LUT from there? The more careful and calibrated each step is, the better, obviously. :slight_smile:

1 Like

Maybe film manufacturers provide response curves for their film types?

@Leonidas & @Morgan_Hardwood

1 Like

As @Morgan_Hardwood explained, i would also shoot an IT8 target and get the photo processed. Best take several shots with different exposures to make sure you have the closest one available in the end. However, instead of manually using curves I would use darktable’s chart tool to create a darktable style from the scanned photo and the reference data of the color target and then use that to process an identity HaldCLUT in darktable. The result should be usable elsewhere.

TL;DR: If you shoot the color target and give me the scans together with the reference file of the target I will have a try at creating a CLUT for you.

About what target to use, the Wolf Faust are said to be good and cheap. However, they require a little care when shooting as they are a bit glossy. For camera profiling the C1 target should be the one you need.

2 Likes

Thanks, I’ll try to aquire a C1 target and shoot the images. Might take a while though, because I’ll wait till my initial film is developed to make sure my camera is not busted (and producing decent images).

@Leonidas
Have you tried Pat David’s APX100 film emulation?
Did you like it? Or what was missing?

I did! My favorite emulation is Rollei Retro 80s though which is said to be very close to APX100 but the emulation of the APX100/Retro 80s looks rather different.

I started this thread partly because I have missed that there is already a APX 100 emulation (:open_mouth:) and partly because I am interested how to make such an emulation in the first place. Kinda learning by doing. Maybe having these color targets shot will allow us to check or improve accuracy of the emulations but that’s admittedly a lofty goal.

The reason I picked APX because that’s the only analog b/w film I can easily buy at many drugstores, so that’s just a pragmatic choice for a starting point.

@houz, do you mean this chart tool? (heh, the one image looks familiar).

And if you follow that link, you will find Fifty Shades of Pat.

Yes, that’s it. It is in development versions of darktable and will be in 2.2.0 once that got released.